Re: Update on Geant4
Moin Stephan,
I have been a long time user of Geant4 in my previous workplace, and
still occasionally lurk around the chat rooms to see what is going on
there. I was made aware that you are trying to get Geant4 into Debian,
and since I consider that a worthwhile effort, I decided to look into
your segfault issues as well.
To make it short, I was not able to reproduce any segfaults with my
compiled version 10.7.0 on my bullseye machine. I downloaded Geant4
straight from the Cern Geant4 Gitlab, and used the same compilation
options I found in your debian/rules file (i.e. using the new option
GEANT4_USE_TBB=1 additionally). The examples I tried, as well as my own
old simulations, still work fine. I also replaced my RunManager with the
new task based RunManager, and that just lead to a huge amount of RAM
being used, but no segfault (it probably caches all events, and I have
to write a map/reduce function to process the results?). I have not yet
checked how the task-based approach is intended to be used, I just
confirmed that it doesn't segfault.
While looking into this, I stumbled upon this presentation:
https://indico.cern.ch/event/942142/contributions/4027521/attachments/2108316/3545985/G4Tasking-Status-and-Overview.pdf
(which is where I found out how to properly instantiate the different
RunManagers nowadays), which mentions on slide 10 something about the
program segfaulting when TBB is used and the G4RunManager is destroyed,
but that didn't happen in my tests, either.
I sadly have no clue how to use the files you put on salsa to build my
own package. I am merely a Debian user, not a developer or maintainer,
but I will look into it how to create a .deb package from those files
(probably dpkg-buildpackage?). If I succeed, I can give you more feedback.
kind regards,
Christoph
--
Jan Christoph Terasa Raum 406, Leibnizstraße 11, 24098 Kiel, GERMANY
Extraterrestrische Physik Telefon: +49 431 880 2487
Institut für Experimentelle und Angewandte Physik www.ieap.uni-kiel.de
Christian-Albrechts-Universität zu Kiel terasa@physik.uni-kiel.de
Reply to: